Structural Dynamics of a Lytic Peptide Interacting with a Supported Lipid Bilayer

Abstract

The interaction of a melittin mutant with a 1,2-dipalmitoyl-sn-glycero-3-phosphatidylcholine (DPPC)-supported lipid bilayer was studied with the use of time-resolved evanescent wave-induced fluorescence spectroscopy (TREWIFS) and evanescent wave-induced time-resolved fluorescence anisotropy measurements (EW-TRAMs). The mutant peptide was labeled at position K14 with AlexaFluor 430 and retained the lytic activity characteristic of native melittin. The fluorescence decay kinetics of the conjugate was found to be biexponential with a short-lived component, τ₁, due to photoinduced electron transfer between AlexaFluor 430 and proximal side chains within or between the peptides. The longer-lived component, τ₂, was sensitive to the polarity of the microenvironment at or near the K14 position of the peptide. Upon interaction with a DPPC-supported bilayer, the proportional contribution of τ₁ increased, indicating a conformational change of the peptide. The values of τ₁ and τ₂ indicate that the AlexaFluor 430 probe experienced an environment with an equivalent polarity no less than that of methanol. EW-TRAMs data from the melittin mutant revealed hindered rotational motions of the AlexaFluor 430 probe both in the plane and perpendicular to the plane of the supported lipid bilayer. The data indicate a highly ordered and polar environment near the center of the melittin helix consistent with the formation of a toroidal pore.

Introduction

There is currently great interest in the application of a class of antimicrobial peptides that includes melittin (1), magainins (2), protegrins (3), cecropins (4), and alamethicin (5) as therapeutics (e.g., antiinflammatories and antibacterials (6,7)), as well as in the treatment of cancers and tumors (8,9). The antimicrobial activity of these peptides and thus their potential for the above applications are linked to their structure. Upon association with cell membranes, these peptides adopt an α-helical conformation whereby the hydrophilic and hydrophobic residues of the peptide chain are sequestered on either side of the helix. The resulting amphiphilic nature of the α-helix grants these peptides detergent-like properties that allow the peptides to disrupt the lipid bilayer of cell membranes, often resulting in cell death.

Since it was first discovered, melittin, the major active peptide constituent of honeybee venom, has become one of the most widely studied peptides within this class of antimicrobial peptides. Its cytolytic activity is based on either transmembrane pore formation (10), resulting in osmolysis, or the breakdown of the lipid bilayer into micelles, which occurs at high peptide/lipid ratios (11). Melittin has been found to adopt different locations, orientations, and association states within membranes depending on experimental conditions and membrane composition (12,13), and thus a range of models have been proposed to explain the mechanism by which melittin lyses cells (14). Despite this body of literature, there is still conjecture about the precise nature of lipid-peptide interactions for melittin and similar cytolytic peptides, as well as the factors that affect these interactions.

Techniques such as oriented circular dichroism (CD) spectroscopy (15,16), neutron diffraction (17), x-ray diffraction (18), and solid-state nuclear magnetic resonance spectroscopy (19) are often used to investigate peptide-lipid interactions. Such methods provide information on the orientation, location, mobility, and conformation of the peptide, and how it associates with the membrane. Complementary information has been provided by fluorescence techniques such as red-edge excitation shift spectroscopy (20), fluorescence quenching (21,22), and time-resolved fluorescence anisotropy measurements (TRAMs) (23,24). However, each of these methods has limitations in studies of interfacial phenomena. Furthermore, bulk solution systems that use lipid vesicles as the membrane structure can entail further complications due to the potential existence of a number of lipid-peptide complexes (i.e., bilayer discs) (25).

One way to overcome these limitations, particularly for bulk solution measurements, is to use techniques that are capable of probing processes close to an interface. To this end, investigators have developed spectroscopic techniques that utilize the phenomenon of total internal reflection. These techniques, which range from Fourier transform infrared-attenuated total reflection (26,27) to evanescent wave-induced fluorescence spectroscopy (28,29), have been used in a number of studies to monitor processes that occur within close proximity to interfaces in a range of different systems. In techniques utilizing total internal reflection, the evanescent (standing) wave, which is produced in the optically rarer medium after total internal reflection of light at an interface between two media of different refractive indices, selectively samples chromophores within the depth of penetration (generally ∼10–100 nm) of the evanescent field. A fluorophore, which acts as a probe that partitions into the interfacial region or is chemically grafted to a surface-specific species of interest, is photoexcited by the evanescent field, and the fluorescence spectrum is collected. Time-resolved evanescent wave-induced fluorescence spectroscopy (TREWIFS) measurements have provided fluorescence decay times, τ_f, of fluorescent species near an interface (30,31). The τ_f-value can report on the polarity of the microenvironment of the fluorophore, from which information on the structure of the interface or the behavior of surface species can be inferred. When time-resolved anisotropy measurements (TRAMs) are incorporated into evanescent wave-induced fluorescence (EW-TRAMS) (32–34), one can monitor fluorescence depolarization events both parallel and perpendicular to the plane of the surface, and obtain information on the mobility of the fluorescent probe or fluorescently labeled species in both of these planes.

Here, we present the first TREWIFS and EW-TRAMs results (to our knowledge) regarding lipid-peptide interactions, specifically the interaction of a fluorescently-labeled melittin mutant with a 1,2-dipalmitoyl-sn-glycero-3-phosphatidylcholine (DPPC)-supported lipid bilayer (SLB) (35). Previous fluorescence-based studies on the interaction of melittin with membranes used fluorescence from the intrinsic tryptophan residue or fluorescent labels attached to the N-terminal amine, lysine-7, or C-terminal residues (13,36,37). The location of the fluorescent probe near the N- and C-termini limits its ability to report on processes occurring at the middle of the helix. In this study, we avoided this limitation by substituting the proline-14 residue with lysine to which the fluorescent label AlexaFluor 430 (Invitrogen, Carlsbad, CA) (38) was attached. The labeled lysine residue lies near the middle of the peptide chain, close to the plane that separates the hydrophilic and hydrophobic regions of the amphiphilic helix. At this location, the AlexaFluor 430 label protrudes from the peptide helix, where it can sample the peptide's microenvironment. Using TREWIFS and EW-TRAMs, we measured the fluorescence lifetime, τ_f, and the fluorescence depolarization dynamics, respectively, of the AlexaFluor 430-labeled melittin mutant during interaction with a DPPC SLB. The τ_f-value reports on the microenvironment of the probe and thus provides insight into the location of the peptide within the bilayer. Fluorescence depolarization is interpreted in terms of the mobility of the peptide when associated with the SLB. The combined interpretation of these two pieces of data provides insight into peptide-lipid bilayer interaction in the context of the proposed models.

Materials and Methods

Cleaning methods

The methods used to clean the glassware and optical components used in this study are described in full in the Supporting Material.

Materials and solution preparation

All general chemicals used in this study and preparations of the buffer, peptide, probe, and probe-labeled peptide solutions are described in the Supporting Material.

TREWIFS techniques

Experimental details about the TREWIFS apparatus (constructed in-house), technique, and incorporation of EW-TRAMs have been described previously (32–34). See the Supporting Material for further details.

Time-correlated single-photon counting (40) was used to measure fluorescence decay profiles for both the bulk solution and TREWIFS measurements. Fluorescence decay profiles were collected to obtain a total of 10,000 counts in the channel of maximum intensity. See the Supporting Material for details.

For EW-TRAMs, fluorescence decay profiles were collected for four excitation-emission polarization permutations (30 min in each excitation-emission permutation), I_ab, where a is the polarization orientation of excitation light and b is that of the emission: I_ss(t), I_sp(t), I_ps(t), and I_pp(t). Using the interface as the frame of reference, s-polarization is in the plane of the interface and p-polarization is normal to the interface. From these data, anisotropy decay functions can be modeled to extract the corresponding rotational correlation times (32). Details are presented in the Supporting Material. The laser power and photostability of AlexaFluor 430 were carefully monitored, and no detectable changes in either parameter were observed over the course of the experiments.

Preparation of the peptide/SLB system

We prepared SLBs using the vesicle deposition method, in which small unilamellar vesicles (SUVs) are adsorbed from solution onto a solid support where they subsequently undergo rupture and fusion to form an extended phospholipid bilayer (35). In this study, we formed DPPC SLBs on the flat face of the fused silica hemicylindrical prism for direct use in TREWIFS and EW-TRAMs studies by incubating a suspension of DPPC SUVs with the hemicylindrical prism at room temperature for 3 days. Quartz crystal microgravimetry showed that this time was required for complete SLB formation (data not shown). The cell was then flushed through with HEPES buffer solution to remove any unadsorbed SUVs. The peptide was then introduced to the cell and allowed to equilibrate overnight before the TREWIFS and TRAMs studies were performed. In all experiments the peptide/lipid ratio was 1:1. The preparation methods used for the DPPC SUVs are described in the Supporting Material.

Peptide synthesis and fluorescent labeling

Melittin (molecular mass: 2847.5 Da) is a 26-residue peptide with the sequence GIGAVLKVLTTGLPALISWIKRKRQQ. It is cationic with a net charge of +6 (a single positive charge on the N-terminal amino group and five positively charged residues) at physiological pH. To form the melittin mutant, proline-14 (P14) was substituted by lysine (K14) to obtain the sequence GIGAVLKVLTTGLKALISWIKRKRQQ. This substitution allows for amine coupling of the fluorescent label, AlexaFluor 430. The procedures used for synthesis of the peptides, selective deprotection, labeling of lysine-14 with AlexaFluor 430, and purification are described in the Supporting Material (41,42).

Lytic assay of the AlexaFluor 430-labeled melittin mutant

We assayed the lytic activity of both native melittin and the melittin mutant by measuring the fluorescence of a self-quenching fluorescent dye encapsulated inside multilamellar vesicles (43). We found that the mutant melittin was nearly twice as effective at releasing dye from the multilamellar vesicles. Full details are given in the Supporting Material. This increase in activity is similar to that found previously for a Pro-14→Ala substituted melittin analog (44).

Environment accessibility studies (fluorescence quenching)

We performed fluorescence quenching studies by measuring the fluorescence emission intensities of free AlexaFluor 430 and the AlexaFluor 430-labeled melittin mutant as a function of the concentration of acrylamide (from 4 mM to 40 mM) in HEPES buffer solution (see Supporting Material for details). The K_SV for melittin mutant (2.6 M⁻¹) was approximately two-thirds that for free AlexaFluor 430 (3.9 M⁻¹). This indicates that AlexaFluor 430 retains a significant degree of accessibility to the collisional quencher and hence its environment when it is covalently bound to the lysine-14 residue of the peptide. The AlexaFluor 430 label on the melittin mutant is therefore well positioned to sample changes in the microenvironment of the peptide as it interacts with a DPPC bilayer.

Results and Discussion

Environmental sensitivity studies of AlexaFluor 430 and the melittin mutant

We characterized the environmental sensitivity of the AlexaFluor 430 probe by performing fluorescence lifetime measurements of the free probe in solvents of different polarity. The resulting fluorescence decay kinetics data are shown in Fig. 1. The fluorescence decay kinetics of free AlexaFluor 430 are monoexponential and, as expected, depend on the polarity of the solvent. The fluorescence lifetime data are summarized in the Supporting Material, together with the $E_T^N$ parameter (a normalized scale of solvent polarity) (45) and the goodness of fit, χ². The time-dependent fluorescence of AlexaFluor 430 decays most rapidly in water. This is unaffected by the presence of HEPES buffer. In lower-polarity solvents, the fluorescence lifetime, τ, is higher than that in an aqueous environment. The low-polarity hydroxylic solvents yield the longest fluorescence lifetimes.

Figure 1.

Fluorescence decay kinetics of both AlexaFluor 430 and the AlexaFluor 430-labeled melittin mutant in solvents of different polarity.

The fluorescence decay kinetics of AlexaFluor 430 when it is chemically attached to the melittin mutant are biexponential (Fig. 1 and Table 1; residuals for the exponential fits are shown in the Supporting Material). There is a short-lived component, τ₁, in the range of 0.8–1.7 ns, and a longer-lived component, τ₂, that is comparable to the florescence lifetime of free AlexaFluor 430. The polarity dependence of τ₂ is similar to that observed for free AlexaFluor 430. The τ₂-value of the AlexaFluor 430-labeled melittin mutant is therefore a good sensor of the local environment of the peptide at the K14 position along the peptide chain. The existence of the short-lived component, τ₁, suggests a process of fluorescence quenching that is not present in free AlexaFluor 430. The possibility of quenching by resonance energy transfer can be discounted because there are no chromophores that absorb in the region of AlexaFluor 430 emission. However, it is possible that photoinduced electron transfer (PET) occurs (46,47). PET between photoexcited organic fluorophores and tryptophan has been reported previously (46,48). In this study, it is possible that the peptide adopted a conformation that enabled PET between the AlexaFluor 430 probe and intrinsic tryptophan along the peptide chain. The notion that PET served as the quenching mechanism is further supported by the fact that τ₁ decreases with increasing solvent polarity, since PET is known to be dependent on the polarity of the solvent (47).

Table 1.

Fluorescence lifetimes of the AlexaFluor 430-labeled melittin mutant in solvents of different polarity

Fluorescence lifetimes of the melittin mutant in solution
Solvent	$E_T^N$	amp(τ1)	τ1(ns)	amp(τ2)	τ2(ns)	χ2
H2O	1.000	12	1.0	88	4.1	1.23
HEPES buffer	∼1	16	0.8	84	3.6	1.26
Methanol	0.762	2	1.3	98	4.8	1.19
Ethanol	0.654	3	1.7	97	5.0	1.17
Acetonitrile	0.460	15	1.7	85	4.9	1.20
Pyridine	0.302	13	1.7	87	4.2	1.16

The fluorescence decay kinetics is biexponential, with two lifetime contributions: τ₁ and τ₂. amp(τ₁) and amp(τ₂) are the percentage contributions of τ₁ and τ₂, respectively, to the fluorescence decay kinetics; χ² is the goodness of fit to the decay curves; and $E_T^N$ is a normalized value of polarity.

Biexponential fluorescence decay kinetics could arise from two conformations of the melittin derivative. One possibility is that each conformation represents a distinct secondary structure of the peptide: one predominantly helical and the other random coil. This might partially explain why decreasing the solvent polarity changes the proportion of the different lifetime components. It is known that organic solvents can increase the helicity of peptides. Another possibility is that the two distinct lifetime forms represent different conformations of the fluorophore with respect to the peptide. A third possibility is that quarternary interactions can change the proportion of different lifetime forms through peptide-peptide interactions. However, we did not find a simple relationship between the polarity and the fractional contribution. The main conclusion we draw from our time-resolved studies is that the fluorophore lifetime is a probe of solvent polarity and changes in peptide conformation that may accompany peptide-lipid or peptide-surface interactions.

TREWIFS

TREWIFS was used to measure the fluorescence decay kinetics of AlexaFluor 430 at an interface for free AlexaFluor 430 adsorbed at the silica-HEPES solution interface and for the AlexaFluor 430-labeled melittin mutant at both the silica-HEPES solution and SLB-HEPES solution interfaces. The fluorescence decay profiles are shown in Fig. 2 and the extracted decay parameters are given in Table 2. The fluorescence decay kinetics of free AlexaFluor430 is monoexponential with a lifetime of 3.1 ns. This is comparable to the lifetime of free AlexaFluor430 in bulk HEPES solution (3.6 ns). This is a sensible result, given that the silica-HEPES solution interface is highly polar. Of interest, other fluorescent probes, such as malachite green (49) and tetramethyl rhodamine (50), exhibit significant changes in their emission properties when located at the silica-solution interface due to enhancement/quenching. This indicates the value of AlexaFluor 430 as a probe of interfacial environments. Any change in its fluorescence decay kinetics is a measure of a change in the polarity of the probe's environment, and not a consequence of the fact that it is at the surface.

Figure 2.

Fluorescence decay kinetics for AlexaFluor 430 adsorbed from solution onto silica (AlexaFluor 430-silica) and for the AlexaFluor 430-labeled melittin mutant adsorbed on silica (peptide-silica) and when interacting with a DPPC SLB (peptide-SLB). All fluorescence decays were obtained using TREWIFS and recorded for the fluorescent species in situ.

Table 2.

Fluorescence lifetimes of AlexaFluor 430 and the AlexaFluor 430-labeled melittin mutant at an interface in situ

Fluorescence lifetimes from TREWIFS measurements
System	amp(τ1)	τ1(ns)	amp(τ2)	τ2(ns)	χ2
AlexaFluor 430-SiO2	–	–	100	3.1	1.03
Peptide-SiO2	57	0.9	43	3.3	1.17
Peptide-bilayer	40	1.2	60	3.7	1.16

The system indicates the molecule and the surface to which it adsorbs or with which it interacts. The fluorescence decay kinetics of AlexaFluor 430 is single-exponential, whereas that of the peptide is biexponential; therefore, the percentage contribution of the two lifetime components is given. χ² is the goodness of fit to the decay curves.

When the AlexaFluor430-labeled melittin mutant was adsorbed at the silica-HEPES solution interface, its fluorescence decay kinetics were biexponential, as was observed for the peptide in bulk solution. τ₁ and τ₂ were similar under both sets of conditions; however, their relative contributions to the decay profiles were significantly different. The proportional contribution of τ₁ increased from 16% in bulk solution (Table 1) to 57% at the silica surface (Table 2). This is reflective of the increase in the number of peptide molecules undergoing PET, which can only be brought about by a conformational change in the peptide due to its adsorption at the interface. Unfortunately, however, the data do not allow us to specify the nature of the conformational change, which could be a change in secondary structure, a change in the distribution of side chains due to specific interaction with charged species on the silica surface, or some combination of the two. High local concentrations of the peptide when adsorbed to silica might also increase the amount of PET due to peptide-peptide associations.

When the AlexaFluor430-labeled melittin mutant interacted with a DPPC SLB, its fluorescence decay kinetics remained biexponential (Fig. 2). The proportional contribution of τ₁ to the fluorescence decay profile increased in comparison with its contribution in bulk solution, indicating a conformational change of the peptide due to its adsorption to/interaction with an interface as discussed above. However, the values of τ₁ and τ₂ both increased (Table 2), indicating that for PET, the probe experienced a lower polarity environment than when adsorbed to silica or in bulk solution. This can only be brought about by partitioning of the peptide into the SLB. However, the changes in τ₁ and τ₂ indicate that the AlexaFluor 430 probe experienced an environment with an equivalent polarity no less than that of methanol (cf. Table 1). This means that K14 of the melittin mutant does not reside in the alkyl region of the bilayer.

EW-TRAMs

When TRAMs are incorporated into evanescent wave-induced fluorescence (32–34), i.e., EW-TRAMs, one can monitor fluorescence depolarization events both parallel and perpendicular to the plane of the surface and obtain information on the mobility of the fluorescent probe or fluorescently labeled species in both of these planes (see Supporting Material for details). In the study presented here, we used this method to monitor the dynamics of the fluorescently tagged melittin mutant at a DPPC SLB surface in situ. These fluorescence anisotropy data are shown in Fig. 3. Fluorescence depolarization processes parallel and perpendicular to the plane of the surface both changed as a function of time, t. This indicates some level of motion of the fluorophore both parallel and perpendicular to the plane of the SLB; however, parallel to the SLB, the change in anisotropy, r(t), and hence the degree of mobility were significantly greater.

Figure 3.

Fluorescence depolarization kinetics of the AlexaFluor 430-labeled melittin mutant interacting with a DPPC SLB. Data were obtained from EW-TRAMs both parallel (in plane) and perpendicular (out of plane) to the SLB. The solid black line is the fit to the data using a simple nonassociative model according to Eq. 3.

The anisotropy profiles can be well fitted by a number of anisotropy models, both associative and nonassociative. For the moment, however, we deal with the simplest models, in which lifetimes and correlation times are nonassociated. We present in Fig. 3 the simplest fit to the data containing a single rotational correlation time, ϕ, the hindered rotator model (51–53), viz:

$$r\left(t\right)=\left(r_0-r_{\infty }\right)e^{-t/\varphi }+r_{\infty }$$ (1)

where r(t) is the anisotropy of fluorescence emission at time t, r₀ is the anisotropy at time zero, and r_∞ is the anisotropy at infinite time.

The fluorescence depolarization parallel to the plane of the surface shows a clear decay over time and thus is easily fitted by Eq. 3 to give a rotational correlation time, ϕ, of 2 ns. It is more difficult to unambiguously fit the fluorescence depolarization data perpendicular to the plane of the surface because the anisotropy function shows little or no change over most of the timescale, indicating very restricted motion in that plane. We do, however, provide a fit to this data set using a ϕ-value that is globally linked to depolarization parallel to the plane of the surface. Note that the limiting anisotropies, r₀ and r_∞, are allowed to vary for each excitation geometry. Of most significance is the comparison between r₀ and r_∞ for fluorescence depolarization in each plane. In the case of unhindered motion and thus complete depolarization, r(t) asymptotes to an r_∞ of zero on a timescale consistent with molecular motion. If molecular motion is completely hindered, r₀ = r_∞. Here, for depolarization in the plane of the surface, r₀ = 0.230 and r_∞ = 0.131, indicating somewhat restricted motion in that plane. However, perpendicular to the plane of the surface, r₀ = −0.239 and r_∞ = −0.219, indicating highly restricted motion in that plane. If the geometrical restriction to motion was provided by an effective cone with hard walls (the wobbling-in-cone model (51)), the cone angles for the in-plane motions would be 24° and 10° for those motions perpendicular to the plane. The presence of nonzero r_∞ could be due to additional motions that were not resolved with our simple model. The timescale of these motions is beyond the reporting capabilities of the fluorophore, so it must be on the scale of tens of nanoseconds or longer. In overview, the good fit to this simple model suggests that the fluorophore is reporting on environments of vastly different rotational mobility (one ϕ of ∼2 ns and one essentially infinite on the timescales probed). Moreover, the extent of motions is geometry-dependent.

Previous anisotropy decay studies of peptide motion in lipid vesicles assigned correlation times in the range of 0.5–5 ns to local probe motions (23,24,54,55) about the point of attachment to the peptide helix. Helix backbone fluctuations were associated with longer correlation times in the range of 30–60 ns (55,56), and rigid-body rotational diffusion for transmembrane peptides was estimated to be on the order of microseconds (57,58). We therefore assign the ϕ-value of 2 ns with local motions of the AlexaFluor 430 probe about its point of attachment to Lys-14. The value is larger than those previously reported for similar-sized probes rotating in the aqueous phase (∼100 ps) (59), but is more consistent with correlation times reported for probes near water-membrane interfaces (23). Our simple nonassociative model does not reveal a significant contribution from helix backbone fluctuations; however, the values of the limiting anisotropies indicate restricted rotational diffusion processes of the probe that may have contributions from helix backbone fluctuations and/or whole-body rotational precession.

It is important to stress the differences between our approach and system compared with previous works. TRAMs on vesicle-peptide systems contain data from all orientations and cannot resolve motions in two distinct planes. Moreover, vesicle-peptide measurements contain contributions from free peptide, whereas EW-TRAMs is predominantly surface-specific. To our knowledge, this is the first observation of geometrically resolvable anisotropy decays from a peptide-lipid system. This observation provides well-defined constraints for possible models of the peptide-lipid interaction and associated dynamics. We next discuss our data in the context of such models.

There are currently three models for the interaction of peptides with lipid bilayers. In the so-called carpet model (60), peptides reside at the membrane-water interface with the long axis of the helix aligned parallel to the membrane surface and the hydrophobic face of the helix facing the membrane. In the two pore models, the helices are aligned with the long axis of the helix traversing the lipid bilayer. In the barrel-stave model (61), the helices form an oligomeric pore, with the hydrophobic portion of the helices facing the acyl chains of the lipid bilayer. In the toroidal pore model, the helices line a lipidic pore whose surface is composed of polar lipid headgroups (62).

The polarity of the local environment, the rotational rate, and the level of rotational restriction experienced by the probe are all consistent with the probe locating in the headgroup region of the lipid bilayer, which is a polar but motionally restricted region of the bilayer (62). This rules out the barrel-stave model, in which the AlexaFluor 430 probe at position L14 would have access to either the very nonpolar acyl chain region of the bilayer or the water-filled pore, where the probe attached to the lysine side-chain would enjoy considerable motional freedom.

If the carpet model were the dominant mechanism for peptide-lipid interaction in our system (at lipid/peptide ratios near unity), one would expect an adsorbed entangled network of peptide molecules on the surface. Some of these molecules would associate directly with the bilayer headgroups that form the SLB surface, whereas others would experience predominantly intra- and intermolecular peptide interactions. This would allow rotational depolarization both parallel and perpendicular to the surface. Previous measurements of rotational diffusion of entangled networks of proteins adsorbed to surfaces indicated more restricted motion parallel to the surface than perpendicular to the surface (34). A complex network of entangled peptides at a lipid surface would also likely provide microdomains experienced by the probe ranging from high polarity (solvent-exposed) to low-polarity microdomains consistent with peptide segment-segment overlap. Our fluorescence lifetime and anisotropy data are inconsistent with these predictions.

This leaves the toroidal pore model, which can account for the polarity of the microenvironment and restricted rotation in both planes. The assignment of a toroidal pore model is supported by CD and oriented CD studies of the melittin derivative in oriented multilayers that reveal a helical conformation and transmembrane orientation at high peptide density (A.C. Rapson, T.A. Smith, B. Wallace, M.L. Gee, and A.H.A. Clayton, unpublished). However, this model does not completely account for the geometry-dependent anisotropy decays, which are unexpected for a single probe protruding from the center of a transmembrane helix into the lipid headgroups of a toroidal pore. Either AlexaFluor 430 has two local conformations and associated rotational environments or there are two populations of peptides. The former possibility is unlikely, because the probe is attached via a linker to a lysine side chain. However, the latter possibility—a mixed carpet/toroidal pore model—could in principle account for the data.

To test whether a mixed peptide-lipid model could account for the EW-TRAMs results, we fitted the anisotropy data using associative models in which each fluorescence lifetime is linked to a particular rotation. Here, the two fluorescence lifetimes of the tagged peptide interacting with an SLB, τ₁ and τ₂ (Table 2), are assigned to distinct populations of peptide, each contributing to fluorescence depolarization in the planes parallel and perpendicular to the bilayer. On the basis of our TREWIFS data, we assign τ₂ to the population of peptides associated with the bilayer headgroups, and τ₁ to the population of peptides whose fluorescence is quenched through PET, as discussed above. r(t) in either plane is the weighted sum of the contribution of these two populations to the net anisotropy (63):

$$r\left(t\right)=r_0\left[{\beta }_1{e}^{-t/{\varphi }_{c1}}+{\beta }_2{e}^{-t/{\varphi }_{c2}}\right]+r_{\infty }$$ (2)

where r₀ is the anisotropy at time t = 0 and ϕ_cx is the rotational correlation time, ϕ_c, of fluorophore j in microdomain x with fluorescence lifetime τ_j. r_∞ allows for the possibility of hindered rotation as in the nonassociative model. β_x is the time-dependent fractional contribution of fluorophores in microdomain x with lifetime τ_jx to the net anisotropy:

$${\beta }_x\left(t\right)=\frac{\sum _{j=1}^N{\alpha }_{jx}{e}^{-t/{\tau }_{jx}}}{\sum _{x=1}^n\sum _{j=1}^N{\alpha }_{jx}{e}^{-t/{\tau }_{jx}}}$$ (3)

where α_jx is the fluorescence lifetime amplitude of fluorophore j in microdomain x. The fits of the associative model to the anisotropy data in the planes parallel and perpendicular to the bilayer are shown in the Supporting Material. The resulting rotation correlation times, ϕ_c, are given in Table 3. Broadly, this model is consistent with the nonassociative model fit. There is a short-lived component associated with a long correlation time (>25 ns) and a long-lived component associated with shorter correlation times (2–4 ns). The association between the short lifetime and long correlation time is consistent with PET between peptides in a oligomeric pore and slow helix fluctuations within a confined environment. The corresponding association between the short correlation time and long lifetime is consistent with surface-associating helices, with the probe experiencing predominantly a polar environment with reduced restriction to rotation.

Table 3.

Correlation times, ϕ_c, for rotational depolarization in the planes parallel and perpendicular to the SLB obtained from an associative model fit to the anisotropy data of Fig. 3

Data from the associative model fit to the time-resolved fluorescence anisotropy
ϕcParallel to bilayer		ϕcPerpendicular to bilayer
Lipid headgroup-associated peptide (lifetime = τ2)	Quenched or self-associated peptide (lifetime = τ1)	Lipid headgroup-associated peptide (lifetime = τ2)	Quenched or self-associated peptide (lifetime = τ1)
2 ns (β2 = 24%)	25 ns (β1 = 76%)	(β2 = 0)	>25 ns (β1 = 100%)

ϕ_c-Values were obtained for two distinguishable populations of peptide: lipid headgroup-associated and quenched or self-associated. β is the fractional contribution of these populations to depolarization in each plane.

Most significantly, the quenched species dominates fluorescence depolarization, with a fractional contribution of 76% in the plane of the bilayer and a 100% contribution in the plane perpendicular to the bilayer. The population of peptides that are lipid headgroup-associated makes no measurable contribution to depolarization perpendicular to the bilayer. This means that there is no rotation of the probe from this population of peptides in that plane. This can only happen if the peptides that line the pores are immobilized and are all similarly oriented. The surface helices that form the carpet only diffuse laterally and are responsible for 24% of depolarization in the plane of the bilayer. This interpretation is consistent with the interpretation of the nonassociative model fit to the data, i.e., a mixed carpet/torroidal pore model for lipid-peptide association in this system. Fig. 4 is a schematic representation of this model. An important point to note is that it is the fluorescence lifetimes in Table 2 that allow us to state that the data point to torroidal pore formation. The associative model anisotropy data presented here highlight the ability of EW-TRAMs to spatially resolve dynamics and hence provide further details to interpret our model of lipid-peptide interaction.

Figure 4.

Schematic representation of mixed carpet/toroidal pore model of peptide-lipid interaction between the AlexaFluor 430-labeled melittin mutant and a DPPC bilayer. The AlexaFluor 430 probe is represented by the shaded hexagon. Illustrated are the two populations of peptide lining the toroidal pore: lipid-associated with the probe residing in the headgroup region of the bilayer, and self-associated with the probe exposed to the solvent but closely located to a neighboring peptide facilitating PET. The lipid-associated helices also form the carpet on the surface of the bilayer.

Conclusions

We have shown that TREWIFS and EW-TRAMs can provide new insights into lipid-peptide interactions by monitoring the fluorescence lifetime and depolarization of a fluorescent probe grafted to the peptide chain. For the model peptide-lipid system studied here, we found that the probe samples microdomains of polarity consistent with it residing in either the polar headgroup region of the SLB or the aqueous solvent. Restricted rotational diffusion was observed both in and perpendicular to the plane of the SLB, indicating a high degree of spatial order of the peptide. These data are consistent with the mixed carpet/toroidal pore model of peptide-lipid interaction.